AI Transformator: Intelligence Day – Autonomous Vehicles

It comes from the future, not to exterminate humanity, as we might fear, but to transform it through artificial intelligence. In the first episode of this new saga, Transformator shows us the route towards autonomous cars.

“I need your clothes, your boots, and your autonomous car.”

Arriving from the future, Transformator is looking for an intelligent transport to get about (as well as for clothes, but that's another story). In this era, vehicles circulate with complete autonomy, without any human intervention at all. This is level 6 autonomy, as defined by the International Organization of Motor Vehicle Manufacturers (OICA). Although we are not yet at this stage in 2019, the step up is not completely science-fiction.

Unable to locate an available autonomous ride share vehicle, Transformator gives up and decides to “3D print” a motorbike to continue his mission to Zenuity. The company that was created jointly by Volvo and Veoneer the pioneer in the Advanced Driver Assistance Systems (ADAS) that cover areas of research into observation and understanding of the road environment, collision avoidance, driver assist and automated parking. The company is developing all the software elements that together will form the artificial intelligence “brain” to drive your car in the future.

More data, more intelligence. More intelligence, more data.

The Zenuity engineers use 30 cars equipped with 12 to 20 ultrasound sensors and 12 cameras, so that the vehicle can perceive the world around it with the utmost precision - generating 4.4 petabytes of data every month.

To ensure they can be driven on their own, they must take on board all the necessary IT infrastructure to ensure no deviations in the complex predictions and calculations as the slightest disturbance to the neural network could have serious consequences. The vehicles are driven autonomously for around eight hours before coming back to the garage, where the data collected is then extracted. A Dell EMC Isilon infrastructure is leveraged to recover and store the vast amounts of video recordings and sensory information.

No time to lose. To avoid being buried under a mountain of continuous information, the model simulations must be performed on at least 25% of the data within 24 hours.

As a new report from Frost & Sullivan, commissioned by Dell Technologies highlights, successful growth will be defined by those that learn to fully utilize and monetize this unprecedented ocean of data. As quoted by Adrian McDonald, "in the digital economy where data is as valuable as oil, no company can afford for this asset to sit unmonetized."

The more mature artificial intelligence technologies become, the closer the autonomous car comes to reality. However, the greater the level of autonomy, the greater the accuracy required and normally the greater the number of sensor systems used. Along with this comes the complexity and difficulty of the fusion of the multiple sources of data in unison.

Between 2014 and 2016, according to McKinsey, the number of ADAS systems rose by over 50% worldwide. The quantity of data produced by these systems is becoming a challenge in terms of analysis, to such a point that this summer, Zenuity signed a partnership with researchers from CERN (European Organization for Nuclear Research) to drive insights from this ocean of data.

Collider working to avoid collisions

“My CPU is a neural net processor; a learning computer”. The CERN teams use FPGAs (Field-Programmable Gate Array) to study the astronomical quantities of data from the LHC (Large Hadron Collider), which is the same type of accelerator used in Dell EMC PowerEdge servers and Dell EMC Ready Solutions for AI infrastructures.

Like particle physics, autonomous cars make extremely rapid decisions that take multiple factors into account. Zenuity therefore hopes that CERN's expertise will help it improve the reaction and anticipation capacities of its systems. “I think it says something important about the collaborative nature of science that an organization like CERN, that conducts high-energy particle collisions, can work with a company that is dedicated to completely eliminating collisions … in traffic”, notes Dr Dennis Nobelius, Zenuity Chief Executive Officer.

The future development of artificial intelligence will be down to a combination of diverse skills and talents. Autonomous cars and the revolution they will cause in the automotive sector are only one example. Transformator has plenty of others to show us. “Follow me, if you want your company to live.”